Regression Kriging Model Research Articles

Bond strength prediction of spliced GFRP bars in concrete beams using soft computing methods

The bond between the concrete and bar is a main factor affecting the performance of the reinforced concrete (RC) members, and since the steel corrosion reduces the bond strength, studying the bond behavior of concrete and GFRP bars is quite necessary. In this research, a database including 112 concrete beam test specimens reinforced with spliced GFRP bars in the splitting failure mode has been collected and used to estimate the concrete-GFRP bar bond strength. This paper aims to accurately estimate the bond strength of spliced GFRP bars in concrete beams by applying three soft computing models including multivariate adaptive regression spline (MARS), Kriging, and M5 model tree. Since the selection of regularization parameters greatly affects the fitting of MARS, Kriging, and M5 models, the regularization parameters have been so optimized as to maximize the training data convergence coefficient. Three hybrid model coupling soft computing methods and genetic algorithm is proposed to automatically perform the trial and error process for finding appropriate modeling regularization parameters. Results have shown that proposed models have significantly increased the prediction accuracy compared to previous models. The proposed MARS, Kriging, and M5 models have improved the convergence coefficient by about 65, 63 and 49%, respectively, compared to the best previous model.

Predicting and Mapping of Soil Hydraulic Properties in Karnataka

Field capacity (FC) and permanent wilting point (PWP) are the two vital hydraulic properties that determine the availability and retention of the water for plant growth. In the present study, profile FC and PWP were mapped in Karnataka using 5352 soil profile datasets. Class pedotransfer function was developed based on soil texture classes, and the total profile FC and PWP were computed for 1 m soil depth. Primary and secondary terrain attributes, vegetation indices, lithology and climatic datasets were used as environmental covariates for the prediction of hydraulic properties. Random forest regression kriging model was calibrated using 80% of total datasets and validated using 20% of datasets. Root mean square error (RMSE), mean error (ME) and the coefficient of determination (R2) were calculated to assess the performance of models. The model recorded RMSE of 127 mm for total profile field capacity (TFC) and RMSE of 87 mm for total profile permanent wilting point (TPWP). The predicted TFC and TPWP in Karnataka are ranged from 58 to 521 mm and 35 to 342 mm, respectively. Available water content (AWC) was calculated as the difference between TFC and TPWP, and the results showed that northern Karnataka has lower AWC (< 50 mm) compared to the south and western part of Karnataka. The present high-resolution (250 m) maps of TFC and TPWP are useful for irrigation scheduling, land use planning and different crop modeling studies.

Soil total carbon mapping, in Djerid Arid area, using ASTER multispectral remote sensing data combined with laboratory spectral proximal sensing data

Spatial quantification of soil attributes is needed to assess and monitor soil resources. Our objective was to map soil total carbon (TC) over 580 ha bare soils, in Djerid arid area (SW Tunisia). One hundred and forty-four soil samples were collected in nodes of 200 m square grid and their spectra acquired in the laboratory at 400–2500 nm have served to radiometrically correct ASTER image using the empirical line method. TC was predicted using partial least squares regression-kriging model, based on the 144 spectra extracted from the nine visible-near infrared ASTER bands. Residual interpolation has improved prediction efficiency. Indeed, PLSR-kriging achieved higher R2 and lower RMSE than PLSR, respectively: 0.78 against 0.53 and 0.52% against 0.16%. Furthermore, spatial distribution of these quantifications has a physical significance. Our results offer relevant method for soil attribute mapping on large scale.

Reducing the Uncertainty of Radiata Pine Site Index Maps Using an Spatial Ensemble of Machine Learning Models

Site Index has been widely used as an age normalised metric in order to account for variation in forest height at a range of spatial scales. Although previous research has used a range of modelling methods to describe the regional variation in Site Index, little research has examined gains that can be achieved through the use of regression kriging or spatial ensemble methods. In this study, an extensive set of environmental surfaces were used as covariates to predict Site Index measurements covering the environmental range of Pinus radiata D. Don plantations in Chile. Using this dataset, the objectives of this research were to (i) compare predictive precision of a range of geostatistical, parametric, and non-parametric models, (ii) determine whether significant gains in precision can be attained through use of regression kriging, (iii) evaluate the precision of a spatial ensemble model that utilises predictions from the five most precise models, through using the model prediction with lowest error for a given pixel, and (iv) produce a map of Site Index across the study area. The five most precise models were all geostatistical and they included ordinary kriging and four regression kriging models that were based on partial least squares or random forests. A spatial ensemble model that was constructed from these five models was the most precise of those developed (RMSE = 1.851 m, RMSE% = 6.38%) and it had relatively little bias. Climatic and edaphic variables were the strongest determinants of Site Index and, in particular, variables that are related to soil water balance were well represented within the most precise predictive models. These results highlight the utility of predicting Site Index using a range of approaches, as these can be used to construct a spatial ensemble that may be more precise than predictions from the constituent models.

Spatiotemporal mapping and assessment of daily ground NO2 concentrations in China using high-resolution TROPOMI retrievals.

Nitrogen dioxide (NO2) is an important air pollutant that causes direct harms to the environment and human health. Ground NO2 mapping with high spatiotemporal resolution is critical for fine-scale air pollution and environmental health research. We thus developed a spatiotemporal regression kriging model to map daily high-resolution (3-km) ground NO2 concentrations in China using the Tropospheric Monitoring Instrument (TROPOMI) satellite retrievals and geographical covariates. This model combined geographically and temporally weighted regression with spatiotemporal kriging and achieved robust prediction performance with sample-based and site-based cross-validation R2 values of 0.84 and 0.79. The annual mean and standard deviation of ground NO2 concentrations from June 1, 2018 to May 31, 2019 were predicted to be 15.05±7.82μg/m3, with that in 0.6% of China's area (10% of the population) exceeding the annual air quality standard (40μg/m3). The ground NO2 concentrations during the coronavirus disease (COVID-19) period (January and February in 2020) was 14% lower than that during the same period in 2019 and the mean population exposure to ground NO2 was reduced by 25%. This study was the first to use TROPOMI retrievals to map fine-scale daily ground NO2 concentrations across all of China. This was also an early application to use the satellite-estimated ground NO2 data to quantify the impact of the COVID-19 pandemic on the air pollution and population exposures. These newly satellite-derived ground NO2 data with high spatiotemporal resolution have value in advancing environmental and health research in China.

A Machine Learning-Based Geostatistical Downscaling Method for Coarse-Resolution Soil Moisture Products

The surface soil moisture (SSM) products derived from microwave remote sensing have a coarse spatial resolution; therefore, downscaling is required to obtain accurate SSM at high spatial resolution. An effective way to handle the stratified heterogeneity is to model for various stratifications; however, the number of samples is often limited under each stratification, influencing the downscaling accuracy. In this study, a machine learning-based geostatistical model, which combines various kinds of ancillary information at fine spatial scale, is developed for spatial downscaling. The proposed support vector area-to-area regression kriging (SVATARK) model incorporates support vector regression and area-to-area kriging by considering the nonlinear relationships among variables for various stratifications. SVATARK also considers the change of support problem in the downscaling interpolation process as well as for solving the small sample size in trend prediction. The SVATARK method is evaluated in the Naqu region on the Tibetan Plateau, China, to downscale the European Space Agency's (ESA) 25-km-resolution SSM product. The 1-km-resolution SSM predictions have been produced every eight days over a six-year period (2010–2015). Compared with five other downscaling methods, the downscaled predictions from the SVATARK method performs the best with in situ observations, resulting in a 24.4% reduction in root-mean-square error with 0.08 m3·m−3 and a 8.2% increase in correlation coefficient with 0.72, on average. Additionally, anomalously low SSM values, an indicator of drought, had a record low anomaly in mid-July for 2015, as noted by previous studies, indicating that SVATARK could be utilized for drought monitoring.

Comparison Quality of Interpolation Methods to Estimate Spatial Distribution of Soil Moisture Content

ABSTRACT Determining and predicting both the spatial and temporal variability of soil moisture content is a critical factor in crop production in semi-arid agricultural systems. The objective of this research was to predict the spatial distribution of soil moisture content using ordinary kriging (OK), regression kriging (RK) and cokriging (CK) models. For that purpose, the soil moisture was monitored in the wet period (March, April and May) and dry period (August, September and October) in Kadınhanı district of Konya Province, Turkey. In each monthly sampling period of the study, a total of 278 samples were collected from locations within an area of 29,608.6 ha. The coefficients of variation (CV) of the soil moisture in dry and wet soil conditions were 0.39 and 0.20, respectively. Soil texture, bulk density (BD) and land use were the most important factors influencing soil moisture. In addition, dry soil generally had a smaller nugget effect than wet soil. Overall, CK was able to better estimate the soil moisture content than OK and RK, as evidenced by the smaller mean absolute error (MAE) and mean square error (MSE). RK estimation is more vulnerable to error than CK and OK estimations because the low coefficients of determination in the regression equation lead to increased error and lower confidence in the estimates. The insights generated by the modeling utilized in this study can assist land use planners, land managers and farmers to more efficiently predict, manage and utilize soil moisture content and hence produce larger crops more sustainably.

Prediction of Soil Depth in Karnataka Using Digital Soil Mapping Approach

Spatial information of soil depth in regional and national level is essential for arriving crop suitability decisions. In the present study, high-resolution (250 m) soil depth map of Karnataka is prepared using digital soil mapping approach. A total of 5174 Soil legacy datasets studied by NBSS&LUP over a period of 30 years is collected and organized for mapping. Quantile regression forest (QRF) and regression kriging (RK) algorithm is tested to predict the soil depth in Karnataka. Topographic attributes derived from digital elevation model, normalized difference vegetation index, landsat-8 data and climatic variables are used as covariates. For model calibration, 80% of soil depth data is used and 20% of data is used for validation. The classical uncertainty estimates such as coefficient of determination (R2) and root mean square error (RMSE) and bias were calculated for the validation datasets in order to assess the model performance. RK model explained maximum variability for prediction of soil depth (R2 = 30%, RMSE = 34 cm) compared to QRF (R2 = 17%, RMSE = 37 cm). Lithology and elevation are found to be most important variables for prediction of soil depth in Karnataka. The predicted soil depth in Karnataka is ranged from 22 to 173 cm, and the present high-resolution (250 m) soil depth maps are useful in different hydrological, crop modelling and climate change studies.

Integrating Ground-based Observations and Radar Data Into Gridding Sub-daily Precipitation

A new and general approach is proposed for interpolating 6-h precipitation series over large spatial areas. The outputs are useful for distributed hydrological modelling and studies of flooding. We apply our approach to large-scale data, measured between 2014 and 2016 at 159 weather stations network of Meteo Romania, using weather radar information and local topography as ancillary data. Novelty of our approach is in systematic development of a statistical model underlying the interpolation. Seven methods have been tested for the interpolation of the 6-h precipitation measurements: four regression methods (linear regression via ordinary least squares (OLS), with and without logarithmic transformation, and two models of generalized additive model (GAM) class, with logarithmic and identity links), and three regression-kriging models (one uses semivariogram fitted separately every 6-h, based on the residuals of the GAM with identity links models, and other two with pooled semivariograms, based on the OLS and GAM with identity links models). The prediction accuracy of the spatial interpolation methods was evaluated on a part of the dataset not used in the model-fitting stage. Due to the good results in interpolating sub-daily precipitation, normal general additive model with identity link followed with kriging of residuals with kriging parameters estimated from pooled semivariograms was applied to construct the final 6-h precipitation maps (PRK-NGAM). The final results of this work are the 6-h precipitation gridded datasets available in high spatial resolution (1000 m × 1000 m), together with their estimated accuracy.

Regional Precipitation Model Based on Geographically and Temporally Weighted Regression Kriging

High-resolution precipitation field has been widely used in hydrological and meteorological modeling. This paper establishes the spatial and temporal distribution model of precipitation in Hubei Province from 2006 through 2014, based on the data of 75 meteorological stations. This paper applies a geographically and temporally weighted regression kriging (GTWRK) model to precipitation and assesses the effects of timescales and a time-weighted function on precipitation interpolation. This work’s results indicate that: (1) the optimal timescale of the geographically and temporally weighted regression (GTWR) precipitation model is daily. The fitting accuracy is improved when the timescale is converted from months and years to days. The average mean absolute error (MAE), mean relative error (MRE), and the root mean square error (RMSE) decrease with scaling from monthly to daily time steps by 36%, 56%, and 35%, respectively, and the same statistical indexes decrease by 13%, 15%, and 14%, respectively, when scaling from annual to daily steps; (2) the time weight function based on an exponential function improves the predictive skill of the GTWR model by 3% when compared to geographically weighted regression (GWR) using a monthly time step; and (3) the GTWRK has the highest accuracy, and improves the MAE, MRE and RMSE by 3%, 10% and 1% with respect to monthly precipitation predictions, respectively, and by 3%, 10% and 5% concerning annual precipitation predictions, respectively, compared with the GWR results.

Factors controlling the spatial distribution of soil organic carbon in Daxing\u2019anling Mountain

Daxing’anling Mountain, in the northeastern part of China, contains a large amount of soil organic carbon (SOC). Using data including topography, climate, and vegetation, the spatial distribution of SOC content was modeled using classical and geography-based statistics, as well as a geographically weighted kriging model. The study findings include: (1) SOC content generally ranges 40–70 g/kg, with high SOC content in the southwest and low SOC content in the southeast; (2) Results of principal component analysis suggested the normalized difference vegetation index is the best predictor of patterns in SOC; and (3) The geo-weighted regression Kriging model well reflects factors influencing spatial distribution of SOC content. This study provides important baseline information for environmental protection in the Daxing’anling Mountain area, as well as general information as to important factors that mediate this important reservoir of soil carbon.

Semiparametric Bootstrapping Forestimating Parameters in Krigingmodel for Deterministic Simulations

In practice, the parameters in the Kriging model are unknown, but they can be estimated based on the behaviour of the observed data. Parameters in the Kriging model can be estimated based on the consideration of Regression-Kriging models. Regression-Kriging models are Universal Kriging models with polynomials of degree zero (Ordinary Kriging), one (Universal Kriging with degree one), or two (Universal Kriging with degree two). A new method for estimating the parameters in the Kriging model with semiparametric bootstrapping is proposed in this paper. The semiparametric bootstrapping procedure works by combining the bootstrap method and Kriging.

Bayesian analysis of the short-term association of NO2 exposure with local burden of asthmatic symptoms in children

Short-term exposure to air pollution has been associated with exacerbation of respiratory diseases such as asthma. Substantial heterogeneity in effect estimates has been observed between previous studies. This study aims to quantify the local burden of daily asthma symptoms in asthmatic children in a medium-sized city. Air pollution exposure was estimated using the nearest sensor in a fine resolution urban air quality sensor network in the city of Eindhoven, the Netherlands. Bayesian estimates of the exposure response function were obtained by updating a priori information from a meta-analysis with data from a panel study using a daily diary. Five children participated in the panel study, resulting in a total of 400 daily diary records. Positive associations between NO2 and lower respiratory symptoms and medication use were observed. The odds ratio for any lower respiratory symptoms was 1.07 (95% C.I. 0.92, 1.28) expressed per 10 μg m−3 for current day NO2 concentration, using data from the panel study only (uninformative prior). Odds ratios for dry cough and phlegm were close to unity. The pattern of associations agreed well with the updated meta-analysis. The meta-analytic random effects summary estimate was 1.05 (1.02, 1.07) for LRS. Credible intervals substantially narrowed when adding prior information from the meta-analysis. The odds ratio for lower respiratory symptoms with an informative prior was 1.06 (0.99, 1.14). Burden of disease maps showed a strong spatial variability in the number of asthmatic symptoms associated with ambient NO2 derived from a regression kriging model. In total, 70 cases of asthmatic symptoms can daily be associated with NO2 exposure in the city of Eindhoven. We conclude that Bayesian estimates are useful in estimation of specific local air pollution effect estimates and subsequent local burden of disease calculations. With the fine resolution air quality network, neighborhood-specific burden of asthmatic symptoms was assessed.

Spatial variability and agglomeration of soil salinity in Minjiang estuary wetland, Southeast China

Understanding the spatial variability and agglomeration of soil salinity is of great significance for the sustainable development of estuarine wetland. Landsat 8 OLI remote sensing image, digital elevation mode and soil surface samples of Minjiang estuary wetland of Fuzhou were used as the data sources. The correlation analysis and principal component analysis were combined to select significant environmental variables and to reduce their dimensions. We analyzed the spatial variability of soil salinity with support vector regression ordinary kriging model (SVROK) and regression kri-ging model (RK), and quantified spatial agglomeration of soil salinity by the spatial autocorrelation analysis. The results showed that three principal components (PCs) extracted by the principal component analysis could explain at least 85% of the total variance in the original dataset and reflected the comprehensive information of vegetation cover, soil properties and topography. Both soil salinity and its residuals were affected by structural factors and random factors. The SVROK model based on principal component (PCs) as input variables can more accurately reflect the spatial variability of soil salinity, with a trend of "higher in the north and lower in the south". The Moran's I of soil salinity was more than 0.5, with significant positive spatial autocorrelation and a higher spatial aggregation degree, displaying the spatial agglomeration characteristics of "high value agglomeration, high value widespread, high value surrounded by low value".

Comparison of additive and isometric log-ratio transformations combined with machine learning and regression kriging models for mapping soil particle size fractions

Digital soil mapping approaches relating to the soil particle size fractions (psf) face the challenge around how to establish the statistical or geostatistical models from large sets of environmental variables, especially in a situation with sparse soil profile data. Recently, many machine learning (ML) models have sprung up with advantages over statistical models. However, few studies focused on the comprehensive comparative analyses between ML and geostatistical models in the soil psf mapping. And the exploration of optimal combination of data transformation and model simulation was even less. Therefore, two transformed methods such as additive log-ratio (ALR) and isometric log-ratio (ILR) transformations combine with two ML models such as boosted regression tree (BRT), random forest (RF) and a classic geostatistical model of regression kriging (RK) were implemented to map soil psf in the Heihe River basin, China. A total of 640 samples and thirteen scorpan factors were collected and used for the comprehensive comparative analysis. Results showed that the scorpan factors such as temperature, precipitation, elevation, soil type, soil organic carbon, vegetation types and normalized difference vegetation index had important impacts on the soil psf mapping. ILR transformation was better than ALR transformation with advantage of improving stability of data distributions and ML models could also improve the mapping performance in comparison with RK models for better handling candidate factors. For these ML models, the RF models had better accuracy performance than the BRT models. In contrast, ILR transformation combined with RF model (ILR_RF) had the best performance, with the lowest root mean square error values (sand, 15.35%; silt, 14.20%; and clay, 6.66%), Aitchison distance value (0.86), standardized residual sum of squares value (0.60), and the highest concordance correlation coefficient value (0.73) and coefficient of determination value (56.69%) for clay content. In addition, ILR_RF had a relatively higher right ratio of soil texture type (68.44%) and better predict performance for most soil texture types. The predicted maps generated from ILR_RF presented more reasonable and smoother transitions. In the future, more ML models should be explored and more variables related to soil psf should be introduced into the models to improve the predictive performance.

Pinna nobilis in the Mar Menor coastal lagoon: a story of colonization and uncertainty

Populations of the Mediterranean fan musselPinna nobilishave progressively decreased over the last decades as a result of anthropogenic activities. The rate of decline has strongly increased since 2016, when a mass mortality event triggered by the parasiteHaplosporidium pinnaeoccurred, and evidence exists thatMycobacteriumspecies may also have played a major role in the event. Indeed, the epidemic has spread throughout the Mediterranean, although coastal lagoons seem to offer a degree of ‘resistance’ against the parasite. In the early 1980s,P. nobilisappeared in the Mar Menor lagoon and rapidly became an important component of the benthos. However, colonization of the lagoon by the fan mussel was cut short in 2016 when a massive mortality event occurred, possibly as a consequence of the environmental collapse that occurred in the lagoon, parallel to the mortality that the species suffered in the Mediterranean that same year. In this study, we estimated the spatial distribution ofP. nobilisin the Mar Menor for 3 periods: 2003-2004, 2013 and 2016. The first 2 periods use published data, and the last period uses data collected in a new campaign. The probability of occurrence for the 3 periods was estimated using random forest and random forest regression-kriging models. The main environmental variables that determined the dispersion and colonization of the bivalve in the lagoon before 2016 are also identified.

Spatio-temporal regression kriging model of mean daily temperature for Croatia

High resolution gridded mean daily temperature datasets are valuable for research and applications in agronomy, meteorology, hydrology, ecology, and many other disciplines depending on weather or climate. The gridded datasets and the models used for their estimation are being constantly improved as there is always a need for more accurate datasets as well as for datasets with a higher spatial and temporal resolution. We developed a spatio-temporal regression kriging model for Croatia at 1 km spatial resolution by adapting the spatio-temporal regression kriging model developed for global land areas. A geometrical temperature trend, digital elevation model, and topographic wetness index were used as covariates together with measurements from the Croatian national meteorological network for the year 2008. This model performed better than the global model and previously developed models for Croatia, based on MODIS land surface temperature images. The R2 was 97.8% and RMSE was 1.2 °C for leave-one-out and 5-fold cross-validation. The proposed national model still has a high level of uncertainty at higher altitudes leaving it suitable for agricultural areas that are dominant in lower and medium altitudes.

Estimating Grassland Carbon Stocks in Hulunber China, Using Landsat8 Oli Imagery and Regression Kriging.

Accurately estimating grassland carbon stocks is important in assessing grassland productivity and the global carbon balance. This study used the regression kriging (RK) method to estimate grassland carbon stocks in Northeast China based on Landsat8 operational land imager (OLI) images and five remote sensing variables. The normalized difference vegetation index (NDVI), the wide dynamic range vegetation index (WDRVI), the chlorophyll index (CI), Band6 and Band7 were used to build the RK models separately and to explore their capabilities for modeling spatial distributions of grassland carbon stocks. To explore the different model performances for typical grassland and meadow grassland, the models were validated separately using the typical steppe, meadow steppe or all-steppe ground measurements based on leave-one-out crossvalidation (LOOCV). When the results were validated against typical steppe samples, the Band6 model showed the best performance (coefficient of determination (R2) = 0.46, mean average error (MAE) = 8.47%, and root mean square error (RMSE) = 10.34 gC/m2) via the linear regression (LR) method, while for the RK method, the NDVI model showed the best performance (R2 = 0.63, MAE = 7.04 gC/m2, and RMSE = 8.51 gC/m2), which were much higher than the values of the best LR model. When the results were validated against the meadow steppe samples, the CI model achieved the best estimation accuracy, and the accuracy of the RK method (R2 = 0.72, MAE = 8.09 gC/m2, and RMSE = 9.89 gC/m2) was higher than that of the LR method (R2 = 0.70, MAE = 8.99 gC/m2, and RMSE = 10.69 gC/m2). Upon combining the results of the most accurate models of the typical steppe and meadow steppe, the RK method reaches the highest model accuracy of R2 = 0.69, MAE = 7.40 gC/m2, and RMSE = 9.01 gC/m2, while the LR method reaches the highest model accuracy of R2 = 0.53, MAE = 9.20 gC/m2, and RMSE = 11.10 gC/m2. The results showed an improved performance of the RK method compared to the LR method, and the improvement in the accuracy of the model is mainly attributed to the enhancement of the estimation accuracy of the typical steppe. In the study region, the carbon stocks showed an increasing trend from west to east, the total amount of grassland carbon stock was 79.77 × 104 Mg C, and the mean carbon stock density was 47.44 gC/m2. The density decreased in the order of temperate meadow steppe, lowland meadow steppe, temperate typical steppe, and sandy steppe. The methodology proposed in this study is particularly beneficial for carbon stock estimates at the regional scale, especially for countries such as China with many grassland types.

Mapping soil organic carbon and total nitrogen in croplands of the Corn Belt of Northeast China based on geographically weighted regression kriging model

Understanding the spatial distributions of soil organic carbon (SOC) and total nitrogen (TN) in croplands is necessary for the sustainable management of soil resources. In this study, a geographically weighted regression kriging (GWRK) model was applied to investigate spatial variabilities of SOC and TN in croplands of the Corn Belt of Northeast China (CBNC). Based on this, a map of the ratio of SOC and TN (C:N) was obtained. Results of geostatistical analysis showed that the best-fit variogram models for SOC and TN were spherical and exponential, respectively; the spatial dependence structures of SOC and TN were strong. The variabilities of SOC and TN were explained by GWRK model having a relatively high R2, and low root mean square error (RMSE) and mean absolute error (MAE) indices. These indices showed the potential applications of GWRK model on a large scale when environmental variables are quite different and the global trend is weak. The area under the curve (AUC) –receiver operating characteristic (ROC) curve exhibited that the prediction of SOC has a general performance whereas that of TN exhibited a good performance. Annual mean temperature had the largest impact on the prediction of SOC and TN in the CBNC. Mediated by topographic and climatic variables, remote sensing-derived variables did not play prominent roles as other studies for the explanation of the distributions of SOC and TN. SOC and TN contents reduced from northeast to southeast of the CBNC, and the average predicted contents were 13.22 g kg−1 and 1.28 g kg−1, respectively. The total level of C:N ratio was relatively low, suggesting high rates of decomposition. Our study results indicated that the influence of human activities on croplands of the CBNC should not be ignored. It also recommended that the improvement of agricultural management was an approach to re-configure the spatial distributions of SOC and TN in the CBNC of China.

Temperature prediction based on a space–time regression-kriging model

ABSTRACTMany phenomena exist in the space–time domain, often with a low data sampling rate and sparsely distributed network of observed points. Therefore, spatio-temporal interpolation with high accuracy is necessary. In this paper, a space–time regression-kriging model was introduced and applied to monthly average temperature data. First, a time series decomposition was applied for each station, and a multiple linear regression model was used to fit space–time trends. Second, a valid nonseparable spatio-temporal variogram function was utilized to describe similarities of the residuals in space–time. Finally, space–time kriging was applied to predict monthly air temperature. Jackknife techniques were used to predict the monthly temperature at all stations, with correlation coefficients between predictions and observed data very close to 1. Moreover, to evaluate the advantages of space–time kriging, pure time forecasting also was executed employing an autoregressive integrated moving average (ARIMA) model. The results of these two methods show that both mean absolute error (MAE) and root-mean-square error (RMSE) of space–time prediction are much lower than those of the pure time forecasting. The estimated temperature curves for stations also show that the former present a conspicuous improvement in interpolation accuracy when compared with the latter.